Two Types of California Central Valley Heat Waves

Transcription

1 Two Types of California Central Valley Heat Waves Virgin River junction with Orderville Canyon UT R. Grotjahn Richard Grotjahn and Yun-Young Lee University of California Davis

2 Outline 1. Introduction Region LSMP Backwards trajectories 2. Two types: Clustering WAF Pacific jet stream 3. Cluster projections 4. Models simulations Projections LSMPs 5. Summary Anna Grotjahn on 5.10 pitch in UT R. Grotjahn

3 1. Introduction California Central Valley (CCV) example events CCV HW extreme events: 3 day minimum majority of valley >95 th % Though ephemeral, they can be important for climate. Can have big impact upon crops, infrastructure, people. Might not show up on monthly means.

4 California Heat Waves LSMPs Extreme heat has large scale meteorological pattern (LSMP) in many variables. Wave train spans Pacific and beyond Temperature LSMP affects the whole US West Coast. Shown: 850hPa heat wave onset Shading is bootstrap significance Yellow >98.5%, Blue < 1.5% LSMPs for N. America in Grotjahn et al. (2015; Climate Dyn.) Grotjahn and Faure (2008, Wea Anal Fcst)

5 Calculating Backwards Trajectories Air parcels that arrive in center of thermal low (red rectangle) are traced backwards in time Trajectories sort into two (or more) groups Average paths over 4 days for 28 events

6 2. Clusters: Two Paths to the Same End Junction of 2 washes in Bryce Canyon UT R. Grotjahn

7 Trajectories => Clusters Cluster analysis refined membership of which group The two groups are color coded. Red: cluster 1: crosses Pacific, strong westerly component Blue: cluster 2: more local, often with easterly component Average paths for 28 events

8 K-means Cluster Analysis Tested K=2, 3, 4. Few members in more than 2 clusters, hence k=2 for this Procedure: divide events into two initial groups based on trajectories nearly all trajectories (all average paths) remain between 500 and 850 hpa. So, examine lower tropospheric composites; identify areas where the two composites have very different properties. select a few area, level, variable, & time to onset combinations where initial groups strongly differ 700 hpa zonal wind at 2 days lead, 600 hpa temperature at 2 days lead, 700 hpa temperature at 1 day lead over 150W-100W, 20N-60N domain. Cluster means: Temperature anomalies; zonal wind anomalies at indicated day before onset.

9 Cluster test Ta control region Are clusters similar at heat wave onset? Are members consistent within each cluster? Average 850hPa Ta at onset in green box for each member (box and whisker) and cluster mean (bar). TA region agree Lee and Grotjahn (2015)

10 Cluster test Ta difference region Do members of one cluster separate from the other cluster? Are members consistent within each cluster? Average 700hPa Ta at 1 day lead in green box for each member (box and whisker) and cluster mean (bar). No overlap.

11 Cluster 1 Cluster 2 WAF Z 500 and WAF 500 WAF from higher latitude in cluster 1, subtropical in other. (WAF = wave activity flux. Similar to a flux of multiple forms of energy. Instantaneous formulation used: Takaya and Nakamura, 2001) Lee and Grotjahn (2015)

12 Differences of the jet stream Cluster 1 Cluster 2 Subtropical jet extension: longer cluster 1 trajectories. Local flow weaker in cluster 2 Jet streak accelerations imply upper level ageostrophic winds convergence (thus sinking below) thick contour: total wind; wind anomalies use shading and thin contours. Lee and Grotjahn (2015)

13 Zion Narrows UT R. Grotjahn 3. Cluster Projections

14 Cluster projections (CP) Different levels than cluster analysis to match model data same cluster members Procedure: 500 hpa zonal wind at 2 days lead, 850 hpa temperature at 2 days and 1 day lead over 150W-100W, 20N-60N domain. (purple) Calculate projections between each event for the 3 field/level/time combinations Average those 3 numbers to get the CP for that cluster Cluster means: Temperature anomaly ( o C); horizontal wind anomaly (m/s)

15 Reanalyses CP scatter plots Red dot for cluster 1 vs blue circle for cluster 2 Mixed (dashed lines) when cluster 1 and cluster 2 differ by < 0.3 Lee and Grotjahn (2015)

16 4. Models Simulations Barrier Canyon UT R. Grotjahn

17 CMIP5 Models studied Model Horizontal resolution HT vs LT CV Grid # Min. Grid # Mean Duration Event # NCEP-NCAR CCSM4 1:288x192 L MRI-ESM1 2:320x160 H bcc-csm1-1-m 2:320x160 L CNRM-CM5 3:256x128 L HadGEM2-CC 4:192x144 H inmcm4 5:180x120 L NorESM1-M 6:144x96 L GFDL-CM3 7:144x90 H GFDL-ESM2G 7:144x90 L GFDL-ESM2M 7:144x90 L bcc-csm1-1 8:128x64 L MIROC-ESM 8:128x64 H MIROC-ESM-CHEM 8:128x64 H FGOALS-g2 9:128x60 L

18 CCSM4 Backwards Trajectories NCEP/NCAR vs NCAR CCSM4 (longitude and latitude ranges differ) Colors assigned from scatter plots

19 Models PC scatter plots Grotjahn and Lee (2015)

20 Model LSMPs Cluster 1 NCEP/NCAR vs 2 popular & lowest resolution models models have weak mid-ocean trough, Ta anomaly onshore Shading: sign counts; 0.7 means 84% of member have same sign Grotjahn and Lee (2015)

21 Model LSMPs Cluster 2 NCEP/NCAR vs 2 popular & lowest resolution models Mid-ocean trough varies; Ta onshore except CCSM Shading: sign counts; 0.7 means 84% of member have same sign Grotjahn and Lee (2015)

22 5. Summary Extreme California heat waves tend to form two ways Composites, trajectories, cluster analyses, all find these two types (plus events that are a mixture) Cluster 1: locally-formed T max; strong westerly flow, extension east of Asian jet; mid/high latitude WAF Cluster 2: expansion southwest of pre-existing max T in SW Canada; weak flow; subtropical WAF Type found is not sensitive to reanalysis nor projection level or area Climate models capture both types, but properties & mix of types varies. Tends to be better for higher resolution

23 Thanks for your attention Questions?